Tuberculosis remains a serious and common disease worldwide. In the U.S. emergence of multi-antibiotic resistant strains, due to the rapid passage of Mycobacterium tuberculosis in patients with AIDS now poses a public health problem. Although the most frequently used vaccine in the world, there is no scientifically-based evidence that BCG prevents primary pulmonary tuberculosis and its protective effect in children has not been related with a protective antigen or a host immune component. Our research into a new vaccine for tuberculosis is based upon the similarity of primary infection caused by M. tuberculosis with that of capsulated bacterial respiratory pathogens, viz: 1) epidemiologic data which show that tuberculous meningitis has a similar age distribution as capsulated bacteria; 2) the presence of a capsular polysaccharide on M. tuberculosis and other mycobacteria in vitro and in vivo; 3) that BCG and protein components of this and wild-type strains, without exception, at best have prolonged but never conferred protection against challenge with wild-type M. tuberculosis in laboratory animals; and 4) published data by F. Seibert that protective immunity in rabbits was correlated with the level of serum precipitating antibodies to M. tuberculosis polysaccharide II proposed to be a linear homopolymer of poly alpha(1 to 2)-D-Glu. This alpha(1 to 2) glucose moiety is identified in extracts of M. tuberculosis, but not in BCG, by its reactivity with pneumococcal type 12F typing antisera. The reactivity in mycobacteria extracted previously with saline and then with 1% Triton. The residual organisms are extracted by the hot phenol method used for LPS of Gram-negatives. The resultant product precipitates with pneumococcal type 12F antisera giving an identity reaction with dextran 1299 which has kojibiose residues. Animal and human antisera, reactive with pneumococcus type 12 and dextran 1299 elicit complement-dependent killing of M. tuberculosis strains Erdmann and wild-type M123. The cidal reaction is inhibited by a synthetic kojipentose but not by an alpha(1 to 4)-D-Gal trimer.